Progesterone together with estrogen orchestrates many cellular responses involved in the regulation of normal female reproductive processes. In its 2 classical target organs, the uterus and the mammary gland, progesterone and its synthetic analogs, progestins, have profound effects on cell proliferation and differentiation. In these organs, progestins can either stimulate or inhibit cell proliferation in a cell type- and tissue-specific manner.1, 2 For example, in the endometrial epithelium, progesterone inhibits estrogen-mediated growth and induces differentiation and is therefore combined with estrogen in postmenopausal hormone replacement therapy (HRT) to control the estrogen-induced cell proliferation. The effect is the opposite in the stromal cells, where progesterone acts synergistically with estrogen to stimulate proliferation. Progestins are also known to inhibit the progression of endometrial cancer.3
In mammary gland, the role of progesterone in controlling the epithelial cell growth is still a matter of discussion. Progestin has a biphasic action on the cell cycle, since it both stimulates and inhibits breast cancer cell proliferation in vitro.1, 4 In these cells, progestins initially stimulate G1 cells to enter S phase, but the prolonged progestin treatment is marked by growth inhibition of the cells. In normal breast, progestins stimulate cell proliferation in vivo;5 in breast tumors and normal breast cells in culture,6 they can inhibit estrogen-stimulated growth, which is often associated with increased differentiation. Progestins are therefore used to treat certain hormone-responsive cancers.2, 7 It has been shown that the application of progestins in HRT might increase the risk of breast cancer,5, 8, 9, 10 but the issue remains controversial.
The growth-inhibitory effects of progestins on the cell cycle in mammary gland have been associated with the decreased expression of cylin D1 and E,11, 12 while stimulatory effect has been explained by an induction of cylin D1 expression.13 Recenly, it was demonstrated that regulation of cyclin D1 is critical for progestin inhibition in breast cancer cells and that overexpression induces progestin resistance in these cells.14 Progestins increase the expression of TGFα, EGF receptor, c-fos and c-myc, but inhibits the activity of CDK-cyclin complexes.1, 12 It should be noted that the regulation of cyclin D1 is not likely to be direct progestin effect, and the molecular basis of progestin effects on other cell cycle genes, such as c-fos, c-myc, c-jun and jun-B, is not well understood.15
Progestin action is mediated primarily through the progesterone receptor (PR), which functions as a ligand-inducible transcription factor. Upon activation by ligand, PR homodimerizes and affects gene transcription by binding to progesterone response elements (PRE) in promoters of target genes. In this way, progesterone presumably triggers the expression of a largely undefined set of genes, which, in turn, leads to activation of further downstream targets. In recent years, a number of articles have appeared describing progestin target genes in breast cancer cells. Examples include mostly positive target genes, some of which might affect the transcriptional activity of PR or play a role in differentiation of breast cancer cells or are involved in membrane-initiated signaling.16, 17
Although some progress has been made in recent years in our understanding of the molecular action of progesterone, the number of identified primary progesterone target genes remains relatively small, since most of the reported genes are not directly regulated by progesterone but require intermediary de novo protein synthesis. Identification of primary molecular targets for progesterone action is of special importance in understanding the progesterone regulation of cancer cell growth and differentiation. Therefore, to better understand the molecular mechanism by which cell proliferation and differentiation are regulated by progestins in human breast cancer, we have applied mRNA differential display technique to identify progestin-regulated genes in MCF7 cells. Here we report one of our findings, hDlg5, a gene encoding a member of the MAGUK (membrane-associated guanylate kinase homologue) superfamily of tumor suppressor proteins, to be progestin regulated in breast cancer cells. Furthermore, we demonstrate that hDlg5 is a primary progesterone target gene and the first hormonally regulated member of the MAGUK gene family.
MATERIAL AND METHODS
Dulbecco's modified Eagle's medium with F12 (DMEM/F12), 17β-estradiol, medroxyprogesteroneacetate, cycloheximide (CHX) and dexamethasone (DEX) were provided by Sigma (St. Louis, MO). Fetal bovine serum (FBS), penicillin-streptomycin and insulin were obtained from Gibco BRL (Groningen, The Netherlands). Dihydrotestosterone (DHT) and progesterone were purchased from Merck (Darmstadt, Germany). Schering Aktiengesellschaft (Berlin, Germany) provided R5020 and Roussel Uclaf (Paris, France) provided RU486.
Cell culture and cell growth assay
MCF-7, T-47D and ZR-5-1 cells were cultured in DMEM/F12 supplemented with 5% FBS, penicillin-streptomycin and 10 ng/ml insulin. Before experimental studies, cells were cultivated 2–3 passages in phenol red free DMEM/F12 supplemented with 5% dextran-coated, charcoal-stripped treated FBS, penicillin-streptomycin, 10 ng/ml insulin and 1 nM 17β-estradiol. Then cells were plated in 96-well plates at a density of 3 × 103 cells per well in the experimental medium. Cells were allowed to attach overnight and medium was replaced. After 24 hr, appropriate steroid hormones in 100% ethanol or ethanol were added. The number of cells was measured using the crystal violet-based nuclei staining method.18
RNA isolation and differential display (DD-RT-PCR)
After being treated for 24 hr or 48 hr with MPA or vehicle, cells were harvested from 150 cm2 plates, total RNA was extracted using TRIzol Reagent (Gibco BRL) according to the manufacturer's protocol, incubated with 2 units of Dnase (Promega, Madison, WI) for 30 min at 37°C and used for differential display. Differential display was carried out using the RNAimage Kit (GenHunter, Nashville, TN) according to the manufacturer's instructions but with some modifications. For each RNA sample, 3 reverse transcription reactions were performed, using 0.2 μg DNA-free RNA and one of the 1-base-anchored oligo-dT primers (H-T11A, H-T11C or H-T11G). PCR reactions were performed using 20 different upstream arbitrary primers (AP1-10 and AP46-55 from the RNAimage Kit) in combination with 3 H-T11M primers, in the presence of 0.25 μl α-[33P]dATP (2,000 Ci/mmole). PCR temperature cycling conditions were: 94°C for 30 sec, followed by 40 cycles of 94°C for 30 sec, 40°C for 2 min and 72°C for 30 sec, followed by 72°C for 5 min. For each primer combination, PCR was performed twice with cDNA derived from 2 different batches of RNA and samples were run in parallel lanes on a same 6% denaturing polyacrylamide sequencing gel. After electrophoresis, the gels were blotted on a 3M paper, dried under vacuum on a gel dryer at 80°C for 1 h and exposed to Kodak BioMax MR film (Eastman Kodak, Rochester, NY) overnight. The bands of interest were excised from the dried sequencing gel, and cDNA fragments were reamplified using the same primers as those used for the display but with 5 nucleotide extension in their 3′ ends to enable direct sequencing of cDNA fragments (sequences of primers available upon request). Annealing temperature for reamplification was raised to 44°C from the original PCR conditions. Reamplified PCR products were run on a 4% agarose, 0.5 × TAE gel, gel purified and subjected to nucleotide sequence analysis. Sequencing was performed using ABI Prism system (Perkin Elmer, Foster City, CA) according to the manufacturer's protocol for cycle sequencing on the GeneAmp 9600 (BigDye Terminator Cycle Sequencing Ready Reaction Kit). Some of the PCR fragments were TA-cloned into pGEM T-easy vector (Promega) according to the manufacturer's instructions.
Northern analysis was performed essentially as described.19 In brief, 30 μg of total RNA from MCF-7, T-47D and ZR-5-1 cells cultured in the presence or absence of estrogen and treated with MPA or with vehicle were run on 1% denaturing agarose gel. Occasionally, message size was estimated by comparing its migration to RNA Millenium Marker (Ambion, Austin, TX). The RNA was transferred to Magna nylon membrane (MSI, Westborough, MA) by overnight capillary blotting in 20 × SSC, baked at 80°C for 2 hr and prehybridized for 2 hr at 42°C in 50% formamide, 5 × SSPE, 5 × Denhardt's reagent, 0.5% SDS and 100 μg/ml sheared salmon sperm DNA. Hybridization was performed for 16–20 hr. [α-32]PdCTP-labeled PCR product or the 1.7 kb fragment generated by NotI/HindIII digestion of the 5′ end of the KIAA0583 gene were used as probes. When a reamplified fragment from the DD-RT-PCR was used as a probe, 10 μM corresponding H-T11M primer was added during labeling. Probes were random primer-labeled using Ready-To-Go DNA labelling beads (Amersham Pharmacia Biotech, Piscataway, NJ). The membranes were washed twice with 2 × SSC/0.1% SDS solution for 20 min at 42°C and once with 1 × SCC/0.1% SDS for 15 min at 42°C. The membranes were exposed to Kodak BioMax MS film at −70°C for varying times. The membranes were stripped by pouring boiling 0.5% SDS on the membranes and allowing them to cool to room temperature. The accuracy of loading was then estimated by hybridizing membranes with [α-32]PdCTP-labeled oligonucleotide complementary to 18S ribosomal RNA.
Nucleotide and amino acid homology comparisons were carried out against the GenBank, EMBL and Swiss-Prot databases at the NCBI with the BLAST network service.20 The search for functional patterns of amino acid sequences was carried out with Prosite,21 Pfam22 and Smart23, 24 databases. Analysis of PEST regions was done using the PEST-FIND-program (EMBnet, Austria), the original algorithm being made by M.C. Rechsteiner and S.W. Rogers.25
Quantitative reverse transcriptase-PCR
Quantitative PCR was performed with the LightCycler using SYBR green detection. LightCycler primers for the quantitative detection of the KIAA0583 mRNA were designed using Primer3 web software (S. Rozen, H.J. Skaletsky  Primer3. Code available at http://www.genome.wi.mit.edu/genome_software/other/primer3.html.). Primer sequences for the KIAA0583 mRNA were the following: sense, 5′-aacctcagtggacagaaagac-3′, and reverse, 5′-aataccttcaggagggaca-3′; for PBGD, sense, 5′-aagtgcgagccaaggaccag-3′, and reverse, 5′-ttacgagcagtgatgcctaccaac-3′. Samples contained 100 ng total RNA, DNA master SYBR green I mix (Roche Molecular Biochemicals, Mannheim, Germany) with 3.25 mM Mn(OAc)2, 0.5 μM primer each in a 20 μl volume. The PCR conditions were as follows: 30 sec at 95°C, then 1 sec at 95°C, 7 sec at 55°C, 10 sec at 72°C for 38 cycles, followed by a melting curve analysis. The expression levels were normalized to the levels of porphobilinogen deaminase (PBGD). Relative mRNA abundance was calculated using the PCR efficiencies and the crossing point deviation of a target gene vs. a control as described.26
Identification of hDlg5 as progestin-inducible gene by mRNA differential display
The differential display technique was applied to identify PR target genes that might contribute to the growth inhibition effects induced by progestins in MCF-7 human breast cancer cells. A prominent upregulation of mRNA was detected in response to treatment of cells with 10 nM synthetic progestin MPA. The cDNA fragment, produced by the AP46 and H-T11A primers, approximately 200 bp in size, was excised from the differential display gel, reamplified with modified oligonucleotides as described above and sequenced. Nucleotide sequence analysis of this fragment showed 100% homology to the 3′ end of the previously published partial mRNA named KIAA0583.27KIAA0583 mRNA is almost identical to P-dlg mRNA reported by Nagase et al.,27 but has a 3′ tail extended by 1477 nucleotides and a 5′ end by 1016 nucleotides. Thereafter, the entire cDNA of KIAA0583/hDlg5 has been cloned.28
The reamplified cDNA fragment from the differential display gel was used as a probe in Northern blot analysis to show that it indeed was progestin-inducible. As shown in Figure 1a, Northern analysis revealed strong induction of 2 transcripts of about 9 kb and 4.4 kb in size after 48 hr treatment with 10 nM MPA compared to the cells treated with 1 nM 17β-estradiol only. Also, a third transcript smaller than 18S was detected. No change in expression was detected between cells treated with estrogen and cells grown in the absence of estrogen. Northern blot was repeated with similar results using the 1.7 kb NotI/HindIII fragment of the 5′ end of the KIAA0583 gene as a probe to confirm that the fragment corresponds to the KIAA0583 transcript and not to the shorter P-dlg mRNA alone (data not shown).
hDlg5 is progestin regulated in different breast cancer cell lines
The cell specificity of hDlg5 gene expression and regulation by MPA was studied by hybridizing Northern blots of total RNA isolated from ZR-75-1 and T47D breast cancer cell lines (Fig. 1b,c). Transcript of about 9 kb was expressed in both cell lines, and the expression of the transcript was strongly induced after 48 hr MPA treatment. Northern analysis showed that hDlg5 is also expressed in normal mammary gland tissues (data not shown).
Time course of hDlg5 stimulation
To examine the time sensitivity of hDlg5 induction by progestin, MCF-7 cells were treated with 10 nM MPA or vehicle in the presence of 1 nM estradiol and harvested 2, 6, 12, 24 and 48 hr thereafter. Quantitative real-time RT-PCR analysis showed that the induction of the hDlg5 transcript is an early event. There was an increase in the mRNA level as early as 2 hr after MPA treatment, and levels were increasing through 6 and 12 hr and stayed high at least up to 48 hr (Fig. 2). The maximal induction was observed 48 hr after treatment (about 9-fold). When treated with medium containing 1 nM estradiol only, no induction of the transcript was seen, except at the 48 hr when there was a slight upregulation. Whether this is a true estrogen effect or due to the cells growing to the confluent was not clear.
hDlg5 is a primary progesterone response gene
Since the early induction of hDlg5 would suggest that it is directly activated by PR, MCF-7 cells were treated with MPA in the presence of the protein synthesis inhibitor, cycloheximide (10 μg/ml) for 6 hr and mRNA was analyzed using LightCycler. As shown in Figure 3, cycloheximide treatment did not inhibit MPA-mediated induction of hDlg5 mRNA, suggesting that this gene is a direct progesterone receptor target, since ongoing protein synthesis is not required. Cycloheximide treatment alone elevated basal expression levels of hDlg5 about 2-fold, while in the presence of both MPA and cycloheximide 7.5-fold induction of hDlg5 mRNA was observed. Furthermore, simultaneous treatment of MCF-7 cells with MPA (10 nM) and 10-fold excess of progestin/glucocorticoid antagonist RU486 (100 nM) resulted in inhibition of MPA-induced hDlg5 expression to the level of that induced by RU486 alone (Fig. 4).
Hormonal specificity of hDlg5 stimulation
We also studied the effects of other hormones on hDlg5 expression in MCF-7 cells originally used in the differential display. Cells were grown in the presence of 1 nM estrogen and treated for 24 hr in parallel with 10 nM synthetic progestin R5020 (17α-21-dimethyl-19- norpregn-4,9-diene-3,20-dione), MPA (medroxyprogesterone acetate), dihydrotestosterone (DHT), synthetic glucocorticoid dexamethasone (9-fluoro-11,17,21-trihydroxy-16-methylpregn-1,4-diene-3,20-dione), progesterone and synthetic antiprogestin RU486 (17β-hydroxy-11β-(4-methylaminophenyl)-17α-(1-propynyl9-estra-4,9-diene-3-one). Interestingly, hDlg5 mRNA was strongly upregulated by all 3 progesterone-like compounds (4- to 6-fold) but not markedly by DHT and RU486. A strong upregulation was also detected by dexamethasone (8-fold) (Fig. 5).
hDlg5 is a member of MAGUK family of proteins with 4 PDZ domains
BLAST homology search, Prosite, Pfam and Smart database analysis for domain profile revealed KIAA0583 to be a member of the growing superfamily of MAGUK proteins with unique domain features (Fig. 6). Usually, MAGUKs consist of 1 src homology 3 domain (SH3), domain homologous to the yeast guanylate kinase (GUK) and 1 or 3 PSD95/Dlg/ZO-1 (PDZ) domains. In addition to 3 PDZ domains of P-dlg reported by Nagamura et al., hDlg5 aminoterminus contains 2 additional PDZ domains.29 We note that the first PDZ domain in P-dlg reported by Nagamura et al., having weak homology compared to other PDZ domains (amino acids 640–741), was not identified by Prosite, Pfam or Smart engines and is therefore omitted from Figure 6.
Like its Drosophila counterpart, DlgA,30 hDlg5 contains PEST sequences, motifs that are typical for proteins with a short half-life.31 hDlg5 contains 3 putative PEST sequences (amino acids 489–424, 658–689 and 825–845, having scores +14.92, +7.39 and +7.08, respectively), while the fly homologue (DlgA) contains 1 (score +5.84). Moreover, hDlg5 was found to contain 1 XPPXY motif in the central region (amino acids 524–527). This motif binds to WW domain found in proteins of diverse function, including regulatory, cytoskeletal and signaling molecules.32, 33, 34
To investigate the molecular basis for progestin-induced inhibition of breast cancer cell proliferation, we used mRNA differential display method to detect progestin-induced alterations of gene expression in MCF-7 cells. One of the isolated cDNA fragments showed notable upregulation by MPA in MCF7 cells. Nucleotide sequence analysis of this fragment revealed homology to the previously published sequence KIAA0583. Based on the homology search, KIAA0583 seems to be a member of the growing family of MAGUKs and is therefore the fifth human homologue of the Drosophila dlg gene.28 hDlg5 contains 4 PDZ domains rather than the usual 1 or 3, an SH3 domain and a GUK domain, which are characteristic structures of MAGUKs. In addition, 3 putative PEST sequences, proline-rich region, and 1 XPPXY motif in the central region were found. The large number of functional domains in hDlg5 indicates that it has a potential to interact with several proteins.
hDlg5 was strongly upregulated by all 3 progesterone-like compounds tested, and by dexamethasone and to a lesser extent by DHT and RU486, while simultaneous administration of RU486 inhibited MPA induction of hDlg5 mRNA. The observation that progestin induction was not blocked by protein synthesis inhibitor cycloheximide suggests that hDlg5 is a direct target gene. Since glucocorticoid and progesterone response elements are identical, dexamethasone is supposed to regulate hDlg5 expression via GR. Interestingly, progestins, RU486 and dexamethasone all have inhibitory effects on MCF-7 cell growth. hDlg5 has not been previously associated with progesterone action. Recently, a number of progestin and glucocorticoid target genes were described, but hDlg5 was not among them, due to the fact that the oligonucleotide array used did not carry a hDlg5 probe.35
MAGUKs serve as molecular scaffolds for signal transduction complexes by utilizing multiple protein-protein interaction modules to assemble receptors, adaptor proteins and cytosolic signaling proteins at the cell membrane. The first MAGUK to be identified was the product of the Drosophila tumor suppressor gene lethal(1)discs-large (dlg). The Dlg protein is involved in epithelial cell growth control, junction structure, maintenance of cell adhesion and cell polarity in Drosophila.30, 36, 37 Human homolog of Drosophila Dlg, hDlg is a peripheral membrane protein expressed in a variety of cell types, including epithelia, and localizes to regions of cell-cell contact in human epithelial cells.38 hDlg interacts with a number of cytosolic structural proteins and proteins involved in cell cycle and tumorigenesis. Examples of the previous are protein 4.1, 2 members of the ERM family38, 39 and of the latter PBK (mitotic serine/threonine kinase) and APC, which forms a complex with hDlg that blocks the cell cycle progression from the G0/G1 to the S phase.40, 41 hDlg has also been shown to interact with several oncoproteins, which perturb its function in cell growth control.42, 43, 44 Recently, KIAA0008 was identified as a human cell-cycle regulated homolog of Drosophila dlg that may play a role in breast and colon tumors.45 The identification of several proteins binding to hDlg supports its involvement in cell growth control. KIAA0583 is the fifth human homolog of Drosophila dlg, and the structural similarity it bears with the other family members would predict that it is also an important cell growth regulator.
Epithelial cells are connected by various types of junctions and the formation of cell-cell adhesion contact sites is a prerequisite for epithelial organization and function. During malignant progression, this highly regulated structure and the expression of necessary proteins are distorted. hDlg has been shown to play an important role in the molecular organization of cell-cell contacts and in the maintenance of differentiated epithelial structure. Recently, it was reported that loss of hDlg expression correlates with more undifferentiated cancer cell phenotype and that hDlg protein is stabilized upon increased cell contact.46 It was shown that loss of ability to upregulate hDlg in response to increased cell contact is a vital step during malignant progression.
Our finding now interestingly links the progesterone action to the control of cell-cell contacts. hDlg5 might have a similar role in cancer cells as hDlg, which could partly explain the differentiation and growth inhibition of epithelial cancer cells upon progestin treatment. Upregulation and stabilization of hDlg5 in breast cancer cells by progestins can be an important step in inducing differentiation and preventing malignant growth. Since hDlg5 has multiple protein-protein interaction domains, the ability of progestins to regulate hDlg5 expression presumably affects signal transduction pathways governed by this protein.
In conclusion, we have identified putative tumor suppressor gene hDlg5 as a primary progestin target in breast cancer cells. Our finding supports the notion that MAGUKs have tumor suppressor activity. These proteins are usually localized at cell-cell junctions where they are involved in cell junction organization, tumor suppression and signaling. Therefore, we conclude that hDlg5 potentially has a role in progestin-induced growth inhibition and differentiation of breast cancer cells.
We are indebted to Dr. H. Gronemeyer and Dr. A. Reitmeir for advice. We thank Dr. O. Ohara from Kazuka DNA Research Institute for providing us KIAA0583 cDNA. We also thank Ms. H. Mäkinen and Ms. T. Eskola for excellent technical assistance and Dr. H. Syvälä for invaluable comments.